Quantitative tightening

The unwinding of the asset holdings built up under quantitative easing. It takes two forms. Passive run off is the common one: as bonds mature, the principal repaid to the central bank is not reinvested, the holding falls away on the redemption schedule, and the reserves that were created to buy those bonds are extinguished. Active sales, where holdings are sold into the market before maturity, are far rarer and have been used sparingly, because a seller of size in a market it once supported moves that market on the way out.

The pace is usually set by monthly caps rather than by a target balance sheet size. A cap states the most that will be allowed to run off in a month by asset type, so the actual decline is the smaller of the cap and what happens to mature, and the path is mechanical and published. That predictability is the design intent: run off is meant to proceed quietly in the background while the policy rate does the active work of setting monetary conditions. Where it stops is the part nobody can publish, because the floor is the quantity of reserves the banking system needs to function smoothly, and that quantity is not observable in advance.

The tempting assumption is symmetry, and it does not hold. Evidence that purchases lowered long yields does not imply that run off raises them by a matching amount, for the plain reason that the purchases were announced into stress and in size while run off proceeds slowly into calm, so announcement effects dominate one and flow effects the other. A previous tightening episode also demonstrated the reserve floor the hard way, when overnight funding rates spiked before anyone had judged reserves to be scarce, which is why practitioners now watch money market spreads as the live indicator of how far the process can run. Balance sheet policy and the policy rate can also move in opposite directions at the same time, so a central bank shrinking its holdings is not necessarily a central bank raising rates.